-- Last Updated: Jan-30-13 5:19 PM EST --
The GPS may not be accurate enough to indicate very minor speed difference across very small distances, but the amount of error it introduces in measuring the length of a fixed course of substantial distance is very small. For example, fairly large error in location determination, such as being off by 50 feet at times, hardly matters at all when measuring the distance between two points that are one or two miles apart. Even that much error over a one-mile distance will allow calculation of overall course speed with a precision that is well within 0.1 mph of the actual figure, which should be good enough for most of us (in fact except for extreme errors in opposite direction, thus being additive, precision will be not far off from 0.001 mph, which is MUCH more precise than average people would care about). The greater the distance, the less this kind of error matters.
Also, don't forget that these kinds of errors are not instantaneous random events. If they were, a stationary GPS would indicate sporadic high-speed movements, but that never happens. The error is more like a slow drift, with the error changing slowly enough that a stationary GPS always knows that it is not moving, and by the same token, one that IS moving at a steady speed will not suddenly indicate drastic speed variation (except perhaps when reception has been obstructed for a little while, but that's another issue entirely). Anyway, none of that even matters for overall speed calculations over great distance, and I only mention it to put Willi's earlier complaint in perspective.